1.2.1 Overall market shares per FHV company

Assuming the 12 companies we have information on cover all the FHV in NYC area, we can sum up all the rides for each FHV company over the July-September period and infer the market share of each company over that 3 months period. We can also calculate the market shares for each month.
For clarity, we use the data where the 9 smallest companies are combined.

ss = apply(aggregate[,2:14],2,sum)
out_global = data.frame(Share = signif((ss/sum(ss[-length(ss)]))*100,3))

aggregate$month = month(aggregate$Date) 

temp = aggregate[,c(2:14,16)] %>% group_by(month) %>% summarize_all(list(sum))
out = cbind(out_global, apply(temp,1, function(x) {x[-1]/sum(x[2:13]) })*100)

total_rides = apply(temp[,-c(1,14)],1,sum)
total_rides = rbind(c(sum(total_rides), total_rides))
rownames(total_rides) = "Number of rides"
out_table = rbind(signif(out[-c(1:9),],3))
colnames(out_table ) = colnames(total_rides) =  c("Overall","July", "August", "September")  

datatable(out_table,  options = list(order=list(list(1,"desc")),searching=FALSE,
                   lengthChange = FALSE,
                   lengthMenu = FALSE,
                   pageLength = FALSE,
                   paging = FALSE,
                   info = FALSE), caption = htmltools::tags$caption(
    style = 'caption-side: bottom; text-align: center;', htmltools::em('Market share of the different providers')
  )) %>% formatStyle(0, fontWeight = 'bold')

Yellow Taxis is the main FHV company in NY with more than 80% of the rides being conducted by their drivers.
Uber is increasing is monthly market share over time (+1.2% between July and September) while Yellow Taxis are decreasing theirs (-1.9% over the same period).

datatable(total_rides,  options = list(order=list(list(1,"desc")),searching=FALSE,
                   lengthChange = FALSE,
                   lengthMenu = FALSE,
                   pageLength = FALSE,
                   paging = FALSE,
                   info = FALSE)) %>% formatStyle(0, fontWeight = 'bold')

#datatable(signif(out[13,],3),  options = list(pageLength = 13, order=list(list(1,"desc"))), caption = htmltools::tags$caption( style = 'caption-side: bottom; text-align: center;', htmltools::em('Market share of the different providers'))) %>% formatStyle(0, fontWeight = 'bold')

It is interesting to note the total number of rides over the 3 months period is above 47 millions.

1.2.2 Per week

Here we increase the resolution of the data and focus on a per week analysis, to see if the trend observed on the monthly data is also visible on the weekly data.
As start/end of the weeks do not coincide with start/end of the months, we focus on weeks 27 to 38 (2014-07-06 to 2014-09-27).
As previously, we look at Yellow Taxis, Green Taxis, Uber and we combine the remaining 9 companies.

aggregate$week = format(aggregate$Date, "%U") #week(aggregate$Date)


# head(aggregate)

# table(aggregate$week) # we discard week 26 and 39 as they have less data points

df_byweek= aggregate[,c(11:13,14,17)] %>% filter(week>26 & week<39) %>% group_by(week) %>% summarize_all(list(sum))
# rename week as first day of the week (Sunday)
df_byweek$Date = aggregate$Date [seq(6,89,7)]
  
# wide to long data
df_byweek_long=melt(df_byweek, id.vars=c("week","Date"))
#head(df2)


ggplot(df_byweek_long, aes(x = Date, y=value, group=variable,color=variable)) + geom_line(size=2) +
  ggtitle ("Total number of trips per week") 

We now look at the market share of the FHV companies, calculated per week.

#ggplot(df2, aes(fill=variable, y=value, x=Date)) + 
#  geom_bar( stat="identity", position="dodge")#position="fill")
percent = df_byweek[,2:5]
percent$sum = apply(percent,1,sum)
percent = percent/percent$sum *100
percent = cbind("Week starting on"=df_byweek$Date,signif(percent[,1:4],3))
datatable(percent, options = list(pageLength = 12,searching=FALSE,
                   lengthChange = FALSE,
                   lengthMenu = FALSE,
                   pageLength = FALSE,
                   paging = FALSE,
                   info = FALSE), caption = htmltools::tags$caption(
    style = 'caption-side: bottom; text-align: center;', htmltools::em('Market share of the different providers')
  )) %>% formatStyle('Week starting on', fontWeight = 'bold')

These observations confirm the previously observed trend and show a steady increase of Uber’s market share and a steady decrease of the Yellow Taxis’.

1.2.3 Per day of the week

We now investigate whether the day of the week has an impact on the market shares, since we saw in Section 1.1 that each FHV has its own trend regarding which day of the week they have more customers.

df_byday= aggregate[,c(11:13,14:15,17)] %>% filter(week>26 & week<39) 
df_byday = df_byday[,c(1:5)]%>% group_by(Day) %>% summarize_all(list(sum))

# wide to long data
df_byday_long=melt(df_byday, id.vars=c("Day"))
#head(df2)


ggplot(df_byday_long, aes(x = Day, y=value, group=variable,color=variable)) + geom_line(size=2) +
  ggtitle ("Total number of trips per week") 

We now look at the market share of the FHV companies, calculated per day.

#ggplot(df2, aes(fill=variable, y=value, x=Date)) + 
#  geom_bar( stat="identity", position="dodge")#position="fill")
percent = df_byday[,2:5]
percent$sum = apply(percent,1,sum)
percent = percent/percent$sum
percent = cbind(Date=df_byday$Day,signif(percent[,1:4]*100,3))
datatable(percent, options = list(pageLength = 12,searching=FALSE,
                   lengthChange = FALSE,
                   lengthMenu = FALSE,
                   pageLength = FALSE,
                   paging = FALSE,
                   info = FALSE), caption = htmltools::tags$caption(
    style = 'caption-side: bottom; text-align: center;', htmltools::em('Market share of the different providers')
  )) %>% formatStyle('Date', fontWeight = 'bold')

In Section 1.1 we showed that the Green taxis were having more customers on Saturdays and this increase in the number of rides also leads to an increase in its market shares on Saturdays.